Open Chinese LLM - GSM8K — leaderboard

Metric: Accuracy (%). Source: huggingface.co. 338 models tracked.

Top models

#ModelScore
1Qwen 2 72B Instruct78.7
2Qwen 2 72B75.36
3Meta-LLama-3-Cat-Smaug-LLama-70B72.93
4Llama3-TenyxChat-70B72.71
5Higgs-Llama-3-70B71.65
6Le_Triomphant-ECE-TW371.57
7llama-3-70B-Instruct-abliterated71.34
8dolphin-2.9.1-llama-3-70B70.96
9Llama 3 70B Japanese Suzume Vector V0.170.58
10TW3-JRGL-v270.36
11YiSM-34B-0rn70.36
12free-evo-qwen72B-v0.8-re70.28
13Rhea-72B-v0.570.2
14Smaug-72B-v0.170.13
15YiSM-blossom5.1-34B-SLERP70.13

Interactive version: theaggregate.ai/benchmark?slug=open-chinese-llm-gsm8k · How the rankings work · Data refreshed daily, snapshot 2026-07-22.